Breaking down non-cost barriers to technology adoption is critical - - PowerPoint PPT Presentation

Breaking down non-cost barriers to technology adoption is critical for the transport-energy transformation

International BE4 Workshop

London, UK April 20-21, 2015

David McCollum, Keywan Riahi, Volker Krey (IIASA) Charlie Wilson, Hazel Pettifor (UEA) Kalai Ramea (UC-Davis) Oreane Edelenbosch (PBL) Zhenhong Lin (ORNL)

ADVANCE project

• EU-FP7 project funded for four years (01/2013 – 12/2016) with 5.7 Mio €
• ADVANCE: “Advanced Model Development and Validation for Improved Analysis of Costs and

Impacts of Mitigation Policies”

• Integrated assessment and energy-economy modeling teams:

PIK (DE; REMIND, MAgPIE), IIASA (AT; MESSAGE), PBL (NL; IMAGE/TIMER), FEEM (IT; WITCH), IPTS (EU; GEM-E3, POLES), UCL (UK; TIAM-UCL), UPMF, Enerdata (FR; POLES), ICCS/NTUA (GR; PRIMES, GEM-E3) CIRED (FR; IMACLIM)

• Topical research teams:

DLR (DE; RE integration & resources), UEA (UK; consumer choice) & Utrecht University (NL; energy demand), NTNU (NO; Material flows & LCA)

• International collaborators:
• Non-EU modeling teams: JGCRI (GCAM), NCAR (iPETS), NIES (AIM), RITE (DNE21+)
• Further international expertise: NREL (renewable energy sources), PIAMDDI & EMF (Model

diagnostics & comparison), Simon Fraser Univ. (energy demand)

2

The context of ADVANCE: Exploring transformations

• Whole-systems models - Integrated Assessment Models (IAMs) and E4

models - are central tools for the analysis of climate change mitigation and sustainable development pathways, both globally and nationally.

• A large number of IAM scenarios have been generated over the past few

years, and form an important basis for international assessments like the IPCC AR5, UNEP Gap Report, Global Energy Assessment etc. (~1200 scenarios in AR5 DB)

3

Modelers continue to hone their "map-making" ability

ADVANCE aims to develop a new generation of energy-economy and integrated assessment modeling tools. The goal is to improve the mapping tools in key areas:

• with strategic importance for the assessment of

mitigation pathways

• where substantial

improvements are needed

Source: Wikimedia Commons Source: NASA Source: Wikimedia Commons 4

Key areas for model improvement…

• End-use technologies providing energy services, drivers of energy

demand, and potentials for energy efficiency improvements (WP2)

• Heterogeneity of consumer preferences, and how behavioral changes

affect energy demand (WP3)

• Innovation, technological change and uncertainty (WP4)
• Supply-side bottlenecks: system integration of variable renewable

electricity (VRE), material and energy requirements, infrastructure lock- ins, land-water-energy-nexus (WP5)

5

Objectives of ADVANCE WP3

(Task 3.1: Improving the representation of demand-side heterogeneity in IA and E4 models)

Increase the heterogeneity of consumer groups in IAM transport sectors Better reflect (non-cost) barriers to advanced vehicle adoption in models Quantify the climate policy cost implications of capturing these barriers Understand which policy levers can reduce the barriers over time, by how much, and for whom Draw upon empirical evidence and detailed behavioral studies to inform the modelling New methodologies New answers to novel questions

Participants in ADVANCE WP3, Task 3.1

• Review of empirical micro-studies led by UEA,

supported by IIASA.

• Pioneering models for first implementation of

behavioral aspects done by IIASA (MESSAGE) and PBL (IMAGE).

• Further implementation/model development will be

conducted by UCL (TIAM), FEEM (WITCH), PIK (REMIND), ICCS (GEM-E3), and DNE-21+ (RITE).

Research Questions

• Which consumer/driver attributes can be

incorporated into IAMs in order to improve transport sector heterogeneity and better reflect barriers to technology adoption?

• How are IAM transport scenarios impacted by

these improved representations of behavior and heterogeneity? (w.r.t. technology choice, climate policy costs, etc.)

• What incentives (policy and financial) might help

to nudge consumer/driver behavior in a desired direction?

Modeling Approach

• 1. Disaggregate IAM transport modules so that

LDV demands reflect a heterogeneous set

• f consumers
• 2. Monetize non-cost vehicle purchase

considerations (barriers to technology adoption) by bringing “disutility costs” from a vehicle choice model into IAMs

Frequent Driver Average Driver Modest Driver

Light-Duty Vehicle Consumers/Drivers

Early Adopter Early Majority Late Majority

Urban Suburban Rural Urban Suburban Rural Urban Suburban Rural

… … … … … … … … <= structure repeated =>

Disaggregation of LDV Mode/Demands

Attitude toward technology/risk Settlement Type Driving Intensity 27 consumer groups in total (= 3 x 3 x 3)

% % % % % % % % % % % % % % % km/yr km/yr km/yr

Implement disutility costs from NMNL Model into IAMs

MA3T (Market Allocation of Advanced Automotive Technologies)

a scenario analysis tool for estimating market shares, social benefits and costs during LDV powertrain transitions, as resulting from technology, infrastructure, behavior, and policies Nationwide Model (9 regions in the US) 1458 consumer groups

Source: ORNL & K. Ramea (UC-Davis)

Example Disutility Cost Data

MA3T_ID MA3T_tech_name RUEAA RUEAM RUEAF RUEMA RUEMM RUEMF RULMA RULMM RULMF SUEAA SUEAM 1 Gasoline ICE Conv 0.45 0.00 1.20 0.45 0.00 1.20 0.45 0.00 1.20 0.50 0.03 2 Diesel ICE Conv 5.89 5.17 7.09 6.52 5.79 7.72 7.13 6.41 8.33 5.98 5.21 3 Natural Gas ICE Conv 13.47 9.64 19.78 16.50 12.67 22.81 19.48 15.65 25.79 13.90 9.87 4 Gasoline ICE HEV 1.88 1.44 2.61 1.92 1.48 2.65 1.96 1.52 2.69 1.82 1.41 5 Diesel ICE HEV 3.54 2.80 4.76 5.76 5.02 6.98 7.94 7.20 9.15 3.45 2.75 6 Natural Gas ICE HEV 13.52 9.63 19.92 16.54 12.66 22.95 19.51 15.63 25.92 13.03 9.37 7 Gasoline PHEV10 2.68 2.31 3.34 3.70 3.33 4.36 4.69 4.33 5.36 2.62 2.28 8 Gasoline PHEV20 3.00 2.67 3.61 5.00 4.67 5.62 6.97 6.64 7.59 2.95 2.64 9 Gasoline PHEV40 1.37 1.14 1.91 1.46 1.23 2.00 1.55 1.31 2.08 1.34 1.13 10 Hydrogen ICE 87.43 49.48 149.98 90.46 52.51 153.01 93.44 55.49 155.99 91.72 51.79 11 Hydrogen FC 79.56 45.24 136.13 82.59 48.28 139.16 85.57 51.25 142.13 77.87 44.34 12 Hydrogen FC PHEV10 53.21 27.51 103.30 56.21 30.51 106.31 59.16 33.46 109.26 52.94 27.68 13 Hydrogen FC PHEV20 50.77 26.16 97.13 53.73 29.13 100.10 56.65 32.04 103.01 49.48 25.57 14 Hydrogen FC PHEV40 36.72 18.89 77.32 39.70 21.87 80.30 42.63 24.80 83.23 36.26 18.81 15 EV 100 mile 12.86 10.77 22.15 22.30 18.11 40.88 45.34 34.87 91.79 12.68 10.77 16 EV 150 mile 17.08 11.07 26.46 30.49 18.47 49.25 65.34 35.28 112.25 16.90 11.07 17 EV 250 mile 20.29 10.91 30.40 37.28 18.52 57.50 82.45 35.55 133.00 20.11 10.91 Key: RU (Rural) / SU (Suburban) / UR (Urban) EA (Early Adopter) / EM (Early Majority) / LM (Late Majority) M (Modest Driver) / A (Average Driver) / F (Frequent Driver) Example: RUEAA = Rural + Early Adopter + Average Driver

• etc. for all 27

consumer groups

Units: 1000$/vehicle Year: 2020

These disutility costs would be added to the standard capital costs of vehicles in models (in $/vehicle).

Region: NORTH_AM; Year: 2030; Group: UREMA

Breakdown of Disutility Cost Sub-components

EV charger installation Model availability Range anxiety Risk premium Refueling station availability

EV100 H2FCV

1 2 3 4 5

amount of driving technology attitude refueling/recharging infrastructure vehicle sales/stock

Region: NORTH_AM

EV100 H2FCV

Sensitivity Analyses to Estimate Disutility Cost Sub-components

Breakdown of Disutility Cost Sub-components

EV100

Region: NORTH_AM; Year: 2030; Group: UREMA

500 ppm CO2eq Baseline

Adding disutility costs leads to slower uptake of AFVs

addition of disutility costs addition of disutility costs

with disutility costs without disutility costs with disutility costs without disutility costs

500 ppm CO2eq

Certain consumer groups adopt AFVs much faster

with disutility costs Early Adopters Late Majority

Year: 2030; Group: UREMA

Regional Differences in Disutility Costs

H2FCV

NORTH_AM INDIA+

Cost reduction here is due entirely to lower km/vehicle/yr

* H2 refueling infrastructure coverage and H2FCV diffusion are at 0%.

But…how should perceptions of low tech. diffusion and limited

• infra. vary across

regions? Utilize empirical insights from social influences literature

Comparison of regional results in a 500 ppm CO2eq scenario

Modest Driver

(13,930 km/veh/yr)

Average Driver

(25,860 km/veh/yr)

Frequent Driver

(45,550 km/veh/yr)

Modest Driver

(5,602 km/veh/yr)

Average Driver

(10,400 km/veh/yr)

Frequent Driver

(18,319 km/veh/yr)

NORTH_AM INDIA+

Research Questions

• How are IAM and E4 transport scenarios impacted

by improved representations of consumer heterogeneity/behavior and better reflections of barriers to technology adoption? (w.r.t. technology choice, climate policy costs, etc.)

• What incentives (policy and financial) might help

to nudge consumer/driver behavior in a desired direction?

• How much can be achieved by changing behavior

and preferences?

Expected Findings and Policy Insights

• The inclusion of non-cost barriers to technology

adoption in the decision-making algorithms of models leads to a considerably slower uptake of advance vehicles than under normal model assumptions.

– e.g., in climate policy scenarios, a shift from electricity/hydrogen to biofuels

• If these barriers fail to be removed, climate policy

costs may be markedly higher.

• Policies supporting early-stage infrastructure can

bring down these barriers, while vehicle purchase subsidies can help compensate for them in the early market phase.

Expected Findings and Policy Insights

CO2 reduction CO2 price ($/ton) EV & H2 share EV & H2 subsidy ($/vehicle)

w/o barriers w/ barriers

EV & H2 share Infrastructure Availability

Marginal abatement cost (MAC) curves will likely shift once models better reflect heterogeneity and non-cost barriers to technology adoption. The impact of vehicle subsidies can be analyzed; these will be affected by heterogeneity and non-cost barriers to technology adoption. Policies supporting the development of early- stage recharging/refueling infrastructure can aid the diffusion of new technologies.

Questions? Comments?

Extra slides

References and Documentation

• Kalai Ramea’s (UC-Davis) IEW-2013, IAMC-

2013, and BE4-2015 presentations

• ORNL MA3T website: http://cta.ornl.gov/ma3t/

Source: Zhenhong Lin (ORNL)

• 10000

10000 20000 30000 40000 50000 60000 70000 80000

Gasoline Diesel Hybrid Plug-in Hybrid Fuel Cell Electric

$/vehicle

Model Availability Risk Premium Refueling Charge Refueler Cost Towing Range Anxiety Cost

Urban Early Adopter Moderate driver

• 10000

10000 20000 30000 40000 50000 60000 70000 80000

Gasoline Diesel Hybrid Plug-in Hybrid Fuel Cell Electric

$/vehicle

Model Availability Risk Premium Refueling Charge Refueler Cost Towing Range Anxiety Cost

Components of Disutility Cost (illustrative, 2020)

Rural Late Majority Frequent driver

Source: Kalai Ramea (UC-Davis)

Which dimensions are uncertain, and which are the most important?

Driver Type (km/veh/yr) (Modest / Average / Frequent) Attitude to New Technology (Early Adopt. / Early Maj. / Late Maj.) Settlement Type (Urban / Suburban / Rural) Data availability, quality, uncertainty? Adequate Lacking Adequate Importance of dimension? Very Strong Very Strong Weak

9 (= 3 x 3) consumer groups are enough

Key determinants of disutility costs

disutility costs

EV charger cost Urban / Suburban / Rural splits Early Adopter / Early Majority / Late Majority splits Modest Driver / Average Driver / Frequent Driver splits NG and H2 station and EV-charger availability km/vehicle/yr for M/A/F Drivers All of these things could/should vary by region and over time. Also by scenario.

Workplan Proposal for Task 3.1

Year: 2014 2015 Month: May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Project Month: 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Review of microstudies & Report on microstudies Pioneering implementation in MESSAGE, IMAGE Distribution of disutility cost data to other teams Implementation in TIAM-UCL, WITCH, ReMIND, GEM-E3 Run scenarios based on updated model implementations Multi-model transport paper Deadline for deliverable Work by IIASA Work by other teams Report/paper writing

Deliverable 3.2

Improving the behavioural realism of integrated assessment models of global climate change mitigation: a research agenda (C. Wilson, H. Pettifor, D. McCollum)

• Submitted in Month 19 (July 2014),

instead of originally planned delivery date of Month 30 (~June 2015)

• Now online at: www.fp7-advance.eu
• Derivative papers in preparation;

insights currently feeding into modeling

Deliverable 3.2

• Specific focus on factors influencing

alternative fuel vehicle purchase decisions

• Identifies importance and challenges

for introducing behavioural features into IAMs.

• typology of behavioural features
• synthesis of current modelling

approaches

• empirical basis for behavioural features

(focusing on AFVs)

• discrete choice experiments (n=16)
• social influence studies (n=72)

Motivation & Background

Behavioural Feature Effect size / influence on choice Heterogeneous decision makers

Age high Value orientation medium – low Gender medium Environmental Awareness high - medium Education medium-low

Non-optimising heuristics

Driving practices low

Non-monetary benefits

Refuelling network high CO2 emissions high - medium Range, battery time, warranties high

Risk preferences (discount rates)

Refuelling location high - medium Vehicle range high - medium Fuel savings medium Social influences high - medium

Social influences

Neighbourhood effects high - medium

Contextual constraints

Refuelling density high Refuelling location high Incentives high

How important and/or useful for IAMs are different behavioural features in discrete choice models of vehicle adoption?

Source: Pettifor and Wilson (UEA)